Semiconductor fabrication uses photoresist in ion implantation, etching, and other processing steps. In the ion implantation steps, the photoresist masks areas of a semiconductor substrate that are not implanted with a dopant. In the etching steps, the photoresist masks areas of the semiconductor substrate that are not etched. Examples of the other processing steps include using the photoresist as a blanket protective coating of a processed wafer or the blanket protective coating of a MEMS (micro electro-mechanical system) device. Following the ion implantation steps, the photoresist exhibits a hard outer crust covering a jelly-like core. The hard outer crust leads to difficulties in a photoresist removal. Following the etching steps, remaining photoresist exhibits a hardened character that leads to difficulties in the photoresist removal. Following the etching steps, residue (photoresist residue mixed with etch residue) coats sidewalls of etch features. Depending on a type of etching step and material etched, the photoresist residue mixed with the etch residue presents a challenging removal problem since the photoresist residue mixed with the etch residue often strongly bond to the sidewalls of the etch features.
Typically, in the prior art, the photoresist and the residue are removed by plasma ashing in an O2 plasma followed by cleaning in a wet-clean bath. A semiconductor etching and metallization process of the prior art is illustrated in block diagram format in FIG. 1. The semiconductor etching and metallization process 10 includes a photoresist application step 12, a photoresist exposure step 14, a photoresist development step 16, a dielectric etch step 18, an ashing step 20, a wet cleaning step 22, and a metal deposition step 24. In the photoresist application step 12, the photoresist is applied to a wafer having an exposed oxide layer. In the photoresist exposure step 14, the photoresist is exposed to light which is partially blocked by a mask.
Depending upon whether the photoresist is a positive or negative photoresist, either exposed photoresist or non-exposed photoresist, respectively, is removed in the photoresist development step 16 leaving a exposed pattern on the oxide layer. In the dielectric etch step 18, the exposed pattern on the oxide layer is etched in an RIE (reactive ion etch) process which etches the exposed pattern into the oxide layer, forming an etched pattern, while also partially etching the photoresist. This produces the residue which coats the sidewalls of the etch features while also hardening the photoresist. In the ashing step 20, the O2 plasma oxidizes and partially removes the photoresist and the residue. In the wet cleaning step 22, remaining photoresist and residue is cleaned in the wet-clean bath.
In the metal deposition step 24, a metal layer is deposited on the wafer filling the etched pattern and also covering non-etched regions. In subsequent processing, at least part of the metal covering the non-etched regions is removed in order to form a circuit.
Nishikawa et al. in U.S. Pat. No. 4,944,837, issued on Jul. 31, 1990, recite a prior art method of removing a resist using liquidized or supercritical gas. A substrate with the resist is placed into a pressure vessel, which also contains the liquidized or supercritical gas. After a predetermined time lapse, the liquidized or supercritical gas is rapidly expanded, which removes the resist.
Nishikawa et al. teach that supercritical CO2 can be used as a developer for photoresist. A substrate with a photoresist layer is exposed in a pattern to light, thus forming a latent image. The substrate with the photoresist and the latent image is placed in a supercritical CO2 bath for 30 minutes. The supercritical CO2 is then condensed leaving the pattern of the photoresist. Nishikawa et al. further teach that 0.5% by weight of methyl isobutyl ketone (MIBK) can be added to the supercritical CO2, which increases an effectiveness of the supercritical CO2 and, thus, reduces a development time from the 30 minutes to 5 minutes. Nishikawa et al. also teach that a photoresist can be removed using the supercritical CO2 and 7% by weight of the MIBK. The substrate with the photoresist is placed in the supercritical CO2 and the MIBK for 30-45 minutes. Upon condensing the supercritical CO2, the photoresist has been removed.
The methods taught by Nishikawa et al. are inappropriate for a semiconductor fabrication line for a number of reasons. Rapidly expanding a liquidized or supercritical gas to remove a photoresist from a substrate creates a potential for breakage of the substrate. A photoresist development process which takes 30 minutes is too inefficient. A photoresist development or removal process which uses MIBK is not preferred because MIBK is toxic and because MIBK is used only when a more suitable choice is unavailable.
Smith, Jr. et al. in U.S. Pat. No. 5,377,705, issued on Jan. 3, 1995, teach a system for cleaning contaminants from a workpiece. The contaminants include organic, particulate, and ionic contaminants. The system includes a pressurizable cleaning vessel, a liquid CO2 storage container, a pump, a solvent delivery system, a separator, a condenser, and various valves. The pump transfers CO2 gas and solvent to the cleaning vessel and pressurizes the CO2 gas to supercritical CO2. The supercritical CO2 and the solvent remove the contaminants from the workpiece. A valve allows some of the supercritical CO2 and the solvent to bleed from the cleaning vessel while the pump replenishes the supercritical CO2 and the solvent. The separator separates the solvent from the supercritical CO2. The condenser condenses the CO2 to liquid CO2 so that the liquid CO2 storage container can be replenished.
Employing a system such as taught by Smith, Jr. et al. for removing photoresist and residue presents a number of difficulties. The pressurizable cleaning vessel is not configured appropriately for semiconductor substrate handling. It is inefficient to bleed the supercritical CO2 and the solvent during cleaning. Such a system is not readily adaptable to throughput requirements of a semiconductor fabrication line. Such a system is not conducive to safe semiconductor substrate handling, which is crucial in a semiconductor fabrication line. Such a system is not economical for semiconductor substrate processing.
What is needed is a method of developing photoresist using supercritical carbon dioxide appropriate for a semiconductor fabrication line.
What is needed is a method of removing photoresist using supercritical carbon dioxide appropriate for a semiconductor fabrication line.
What is needed is a supercritical processing system which is configured for handling semiconductor substrates.
What is needed is a supercritical processing system in which supercritical CO2 and solvent are not necessarily bled from a processing chamber in order to create a fluid flow within the processing chamber.
What is needed is a supercritical processing system which meets throughput requirements of a semiconductor fabrication line.
What is needed is a supercritical processing system which provides safe semiconductor substrate handling.
What is needed is a supercritical processing system which provides economical semiconductor substrate processing.